Optimistically Tempered Online Learning
Abstract: Optimistic Online Learning algorithms have been developed to exploit expert advices assumed optimistically to be always useful. However, it is legitimate to question the relevance of such advices \emph{w.r.t.} the learning information provided by gradient-based online algorithms. We develop in this work the \emph{optimistically tempered} (OT) online learning framework as well as OT adaptations of online algorithms. Our algorithms come with sound theoretical guarantees in the form of dynamic regret bounds and we eventually provide experimental validation of the usefulness of the OT approach.
Submission Length: Regular submission (no more than 12 pages of main content)
Previous TMLR Submission Url: https://openreview.net/forum?id=WNa14lmKen&referrer=%5BAuthor%20Console%5D(%2Fgroup%3Fid%3DTMLR%2FAuthors%23your-submissions)
Changes Since Last Submission: The main additional contribution in this revised version is the original framework of Optimistically Tempered Online Learning. We invite the reader to check the cover.pdf file in the supplementary for a detailed survey of the major changes w.r.t. the last version and the incorporation of the feedback from previous reviews.
Assigned Action Editor: ~Nishant_A_Mehta1
Submission Number: 3265
Loading